1

I'm looking at implementing xenserver and would like to know how to get xenserver to gracefully shutdown all vms after a power failure.

3 Answers 3

3

The solutions offered are too simplistic to work in practical situations.

We use the Pacemaker cluster resource manager to manage our virtual machines. Nagios monitors the APC UPSes we have via their network management cards, and runs a script when power failure is imminent to put dom0 servers (also pacemaker nodes) into standby, effectively migrating all resources including virtual machines away from them.

Pacemaker handles the case where resources cannot migrate anywhere relatively gracefully by doing a quick shutdown of the domU machines. As soon as the pcmk node/dom0 machines come back up, the 'resources' (aka domU VMs) get started again where permitted by the pacemaker cluster rules.

2

It appears that apc smartups products work: http://support.citrix.com/article/CTX119910 and http://techblog.deptive.co.nz/2009/04/automated-xenserver-host-vm-shutdown.html and http://forums.citrix.com/message.jspa?messageID=1339540

0

I guess I'm missing the point. Most of these solutions promote a separate Windows machine to run a script.

What assures me: 1) The Windows box doesn't lose power before the XS UPS triggers battery low? 2) If multiple XS hosts, each with separate UPS devices -- do I need a separate Windows machine for each?

Seems like a solutions based on the XS physical machine would be superior. Respond to a signal you're about to be impacted (battery low) -- each pool member need only coordinate with others (within the pool). Theoretically, it is enough to take care of VMs running on your own server and shutdown. (Gets more complicated if the master is already down -- when processing as a pool member (now in emergency mode)).

In my view the "solutions" offered are too simplistic.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .